Metraipolis › archive › dc_d967577fdb69
Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing — arXiv:2610.38076
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.38076
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:11:17.324Z
- SHA-256 of the archived text
93e6f2131e6c040f7c84d18e7a3e5a213e32fc6ae6b899e9d822c4cfb52e1d7b- Stable URL of this page
https://metraipolis.com/documento/dc_d967577fdb69
How to cite
Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing — arXiv:2610.38076 — arXiv (en), https://arxiv.org/abs/2610.38076 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d967577fdb69 (Metraipolis archive dc_d967577fdb69, sha256 93e6f2131e6c040f7c84d18e7a3e5a213e32fc6ae6b899e9d822c4cfb52e1d7b)
BibTeX
@misc{dcd967577fdb69,
title = {Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing — arXiv:2610.38076},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.38076},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d967577fdb69; sha256 93e6f2131e6c040f7c84d18e7a3e5a213e32fc6ae6b899e9d822c4cfb52e1d7b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38076. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM-as-a-Judge has become a standard for evaluating large language models, yet most approaches implicitly assume that a single judge prompt functions as a stable evaluation instrument. Prior work has primarily explored deliberate prompt modifications such as scoring rubrics, role descriptions, or chain-of-thought reasoning in order to optimize accuracy. In contrast, we investigate whether validated prompt variants intended to preserve evaluation semantics while differing in surface realization (\textit{paraphrastic rephrasings}) yield consistent judgments under a fixed evaluation task. We introduce \textit{Prompt Lottery}, a phenomenon in which paraphrastic rephrasings produce different LLMaJ judgments. Across a large-scale study spanning diverse evaluation tasks, judge families, and model backbones, we find that these variations can alter evaluation outcomes and, in some settings, reverse relative model rankings. A user study with 10 human annotators shows that LLMaJ judges exhibit 6-9$\times$ larger agreement degradation under prompt rephrasings than humans, falling below the generally accepted threshold in usability studies. This indicates that single-prompt evaluation is insufficiently robust for model ranking. Finally, we show that multi-prompt aggregation substantially improves robustness by emphasizing judgments that remain consistent across prompt variants.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.38076 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T06:07:08.621Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).